Abstract:

The gravity inversion results of three-dimensional density interface are often not unique, which brings some difficulties to further scientific research. The classical particle swarm optimization algorithm has a higher global extremum search ability, faster inversion speed in computing high-dimensional nonlinear inversion problems, and the final solution is independent of the initial model compared with traditional inversion density interface algorithms such as L-M, Tikhonov regularization, Gauss-Newton method, etc. However, in classical particle swarm optimization, the initial model setting and parameter selection are not perfect. Therefore, this paper further enhances the algorithm based on the classical particle swarm optimization algorithm, referring to the previous optimization ideas. The test results of various models show that the optimized particle swarm optimization algorithm has a stable ability to search for the optimal global solution, and the depth error is smaller. In addition, if we adopt parallel computing, the inversion speed can be effectively improved.

We obtained the Indosinian density interface depth model of the Changning area by inversion using multiple measured high-density gravity profile data based on the improved algorithm. The overall scope of the survey area is small and diamond-shaped, including the complete Changning-Shuanghe anticline and some surrounding synclines. The inversion results show that the Indosinian density interface generally presents the characteristics of uplift in the middle and depressions around it, and the depth range is 0.3~3.3km, which is basically consistent with the inversion results of the drilling data and previous gravity data, and the details are more prominent. It can better express its structural characteristics. The depression degree of the interface on the right side is significantly larger than that on the left side. The uplift part corresponds to the Changning-Shuanghe complex large anticline, and the depth varies from 0.3km to 1.9km. The core of the anticline is exposed to the surface by uplifting and erosion of the tectonic movement. The inversion result provides essential information for studying the seismotectonic environment and is also a vital reference for studying the multi-layer density interface model.

Density interface fluctuation is the product and sign of a specific area under the action of multi-stage tectonic movement, which plays an essential role in studying basin basement, regional structure, and deep structural fluctuation. It provides critical information for the analysis of the origin of earthquakes. Therefore, we analyzed the structural characteristics of this area and its relationship with earthquakes combined with the undulating morphology of the Indosinian surface. Earthquakes in the Changning area are concentrated on the north and south sides of the large anticline. The seismic distribution pattern and focal parameters on both sides are obviously different. The main reason for this phenomenon is that there are significant differences in the causes of earthquakes. The Indosinian surface in the north wing of the anticline is steeper than that in the south wing. The location of the strip distributed shallow earthquakes in the north wing is highly related to the fluctuation of the Indosinian surface, and they mainly occur at the places where the Indosinian surface fluctuates violently. The local density changes drastically, and the earthquakes’ occurrence is greatly affected by hidden faults. The clumped distributed shallow earthquakes in the south wing occur at locations where there is an apparent depression on the Indosinian surface, which may be caused by shale gas exploitation, and the earthquakes are more affected by local stress changes. Deep earthquakes may be closely related to the revival of basement faults. There may still be seismic risk in the northeast wing of the large anticline in the future.

In general, the optimized particle swarm algorithm has achieved good results in both model testing and practical applications. In order to further improve the accuracy of the inversion results, we will focus on improving the applicability of the algorithm in various situations and the ways of adding multiple constraint information. More detailed geophysical research should be carried out in this area, which will help to better understand its crustal structure, earthquake mechanism, geological structure, and the development of earthquake prevention and disaster reduction.

Key words: Changning area, PSO, Indosinian surface, gravity inversion

摘要：

为减少密度界面重力反演的非唯一性, 文中对经典粒子群算法的初值模型和参数设定进行优化, 并通过理论模型验证其有效性。基于该改进算法, 采用长宁科考的多条实测重力剖面数据反演计算了长宁地区印支密度界面的深度, 并结合界面起伏形态分析该地区的构造特征及其与地震的关系。研究结果表明, 经优化的粒子群算法全局搜寻最优解的性能稳定, 深度误差更小。对长宁地区印支密度界面的反演显示, 该界面总体呈现中间隆起、 四周凹陷的特征, 深度介于0.3~3.3km, 与前人的结果基本一致, 文中所得结果的细节更精细。研究区东侧界面的凹陷程度显著大于西侧, 界面隆起部分对应长宁-双河复式大背斜, 深度介于0.3~1.9km, 背斜核部受到构造运动的抬升侵蚀作用导致古老地层出露地表。大背斜北翼处的界面较南翼陡峻, 呈条带状分布的北翼浅层地震发震位置与界面起伏具有高相关性, 地震多发生在印支界面剧烈起伏处; 呈团状分布的南翼浅层地震发震位置处的印支界面存在明显凹陷。文中的反演结果为该地区地震构造环境研究提供了基础信息, 也是后续多层密度界面模型的重要参考。

关键词: 长宁地区, 粒子群算法, 印支界面, 重力反演